$\texttt{SPIN}$: distilling $\texttt{Skill-RRT}$ for long-horizon prehensile and non-prehensile manipulation

TL;DR

This paper introduces SPIN, a framework that distills a complex skill planning algorithm into an efficient policy for long-horizon manipulation, enabling robots to perform complex tasks with high success rates.

Contribution

The paper presents Skill-RRT with skill applicability checks and connectors, and demonstrates zero-shot transfer of the distilled policy to real robots for complex manipulation tasks.

Findings

Achieves over 80% success rate on three long-horizon tasks.

Outperforms state-of-the-art hierarchical RL and planning methods.

Zero-shot transfer from simulation to real-world robots.

Abstract

Current robots struggle with long-horizon manipulation tasks requiring sequences of prehensile and non-prehensile skills, contact-rich interactions, and long-term reasoning. We present $\texttt{SPIN}$ ($\textbf{S}$kill $\textbf{P}$lanning to $\textbf{IN}$ference), a framework that distills a computationally intensive planning algorithm into a policy via imitation learning. We propose $\texttt{Skill-RRT}$, an extension of RRT that incorporates skill applicability checks and intermediate object pose sampling for solving such long-horizon problems. To chain independently trained skills, we introduce $\textit{connectors}$, goal-conditioned policies trained to minimize object disturbance during transitions. High-quality demonstrations are generated with $\texttt{Skill-RRT}$ and distilled through noise-based replay in order to reduce online computation time. The resulting policy, trained entirely in simulation, transfers zero-shot to the real world and achieves over 80% success across three challenging long-horizon manipulation tasks and outperforms state-of-the-art hierarchical RL and planning methods.